2833Research Article

IntroductionOscillatory bending movement is a prominent feature of eukaryoticflagella and cilia. Many sperm flagella, including those of seaurchins, beat with almost planar bending waves. To form a planarbend, dynein arms of the doublet on the two ‘sides’ (with respectto the beat plane) of the central-pair complex (CP) play major roles(Brokaw, 1989a; Holwill and Satir, 1994; Nakano et al., 2003;Shingyoji et al., 1977) (Fig. 1A,B). The doublets of the axonemeare probably all similar in their dynein functions, but the CP isthought to act as a kind of distributor of signals to regulate theactivity of dynein arms according to their position in the axonemeand the phase of beating (Bannai et al., 2000; Nakano et al., 2003;Omoto et al., 1999). Thus, the dynein arms on both sides of the CPare those that are assigned by the CP to produce the microtubulesliding necessary for bend formation. As the dynein molecule is aminus-end-directed motor protein (Sale and Satir, 1977; Vale andToyoshima, 1988), the activity of dynein arms on the two sides ofthe CP should alternate cyclically in order for cyclical bending inalternate directions to occur. How such alternation of dyneinactivity between the two sides of the CP is regulated along theflagellum is therefore important for understanding the mechanismof oscillation.

Several properties of flagella seem to be of particular relevanceto the mechanism of oscillation. Flagella can respond to externallyapplied bending by changing the way their microtubule sliding isregulated (Eshel and Gibbons, 1989; Okuno and Hiramoto, 1976;Shingyoji et al., 1995). In addition, bending an immotile flagellumcan stimulate its dynein to start beating (Hayashibe et al., 1997;Ishikawa and Shingyoji, 2007; Okuno and Hiramoto, 1976). Thus,

it is likely that the flagellar bending itself is involved in theregulation of dynein activity, which would imply that a self-regulatory feedback mechanism underlies flagellar oscillation. Thisidea is strongly supported by our previous study using elastase-treated axonemes of sea urchin sperm, which showed that undercertain conditions externally applied bending causes a reversal ofthe direction of active microtubule sliding (Morita and Shingyoji,2004). Taken together with an ultrastructural study (Nakano et al.,2003), it is tempting to speculate that the mechanical effects ofbending may have initiated ‘switching’ of the dynein activitybetween the two sides of the CP, probably from doublet 7 to doublet3 (or 4) (see below) to induce backward sliding (Fig. 1B,D; pinkand blue bars in Fig. 1C). As the activity of dynein arms on thedoublets other than 7 and 3 (or 4) can also be switched on byexternally applied bending in elastase-treated axonemes (Fig. 1C,grey bars), although independent of the presence of the CP (Moritaand Shingyoji, 2004), we predict that bending regulates the activityof all dynein arms (Fig. 1C).

In this study, as a step to confirm our self-regulatory feedbackmodel (Fig. 1C), we aimed to demonstrate that the switching of thedynein activity between the two sides of the CP does causebackward sliding (Fig. 1D). To do this, we had to solve twoproblems: first, establishing a way to induce backward sliding morestably – in the previous experiments, backward sliding could beinduced by externally applied bending in only about 45% of thefragmented axonemes (Morita and Shingyoji, 2004); second, findingan indicator of dynein activity that could discriminate between theside of the CP with active dynein arms and the side with inactivearms.

Oscillatory movement of eukaryotic flagella is caused by dynein-driven microtubule sliding in the axoneme. The mechanicalfeedback from the bending itself is involved in the regulationof dynein activity, the main mechanism of which is thought tobe switching of the activity of dynein between the two sides ofthe central pair microtubules. To test this, we developed anexperimental system using elastase-treated axonemes of spermflagella, which have a large Ca2+-induced principal bend (P-bend) at the base. On photoreleasing ATP from caged ATP, theyslid apart into two bundles of doublets. When the distal overlapregion of the slid bundles was bent in the direction opposite tothe basal P-bend, backward sliding of the thinner bundle wasinduced along the flagellum including the bent region. The

velocity of the backward sliding was significantly lower thanthat of the forward sliding, supporting the idea that the dyneinactivity alternated between the two sides of the central pair onbending. Our results show that the combination of the directionof bending and the conformational state of dynein-microtubuleinteraction induce the switching of the dynein activity inflagella, thus providing the basis for flagellar oscillation.

Supplementary material available online athttp://jcs.biologists.org/cgi/content/full/121/17/2833/DC1

Key words: Dynein, Flagella, Oscillation, Imposed bending, Sperm,Sea urchin

Summary

Mechanism of flagellar oscillation – bending-inducedswitching of dynein activity in elastase-treatedaxonemes of sea urchin spermShuichi Hayashi and Chikako Shingyoji*Department of Biological Sciences, Graduate School of Science, University of Tokyo, Hongo, Tokyo 113-0033, Japan*Author for correspondence (e-mail: chikako@biol.s.u-tokyo.ac.jp)

Accepted 16 June 2008Journal of Cell Science 121, 2833-2843 Published by The Company of Biologists 2008doi:10.1242/jcs.031195

Jour

nal o

f Cel

l Sci

ence

2834

To overcome the first problem, we need to find the conditionsfor bending to induce backward sliding. Among the possiblevariables, the direction of bending relative to the beat planeseems to be important. In sea urchin sperm, structural relationsin the formation of the principal (P-) and the reverse (R-) bendsat the base of a flagellum are well established. Electronmicroscopy, together with the analysis of bending responsesunder low and high Ca2+ conditions, have shown that the dyneinarms on doublets 7 and 3 (or 4), which are located on two sidesof the CP, are responsible for the formation of the P- and R-bends, respectively (Nakano et al., 2003; Sale, 1986) (Fig. 1A,B).At high concentrations of Ca2+, the formation of R-bends issuppressed owing to a decrease, mediated by the centralpair/radial spoke (CP/RS) system, of dynein activity on doublet3. By contrast, the activity of dynein arms on doublet 7 is notinfluenced by Ca2+ (Bannai et al., 2000; Nakano et al., 2003).As a result, flagella at high Ca2+ are arrested in a so-called‘quiescent’ form that is characterized by a large P-bend at thebase (Fig. 1E). It is interesting that the dynein activity on doublet7 seems to be dominant and that this property is independent ofCa2+ concentration, and thus, under many conditions, the‘quiescence’ is commoner than the ‘relaxed’ straight form (Sale,1985; Yoshimura et al., 2007). This means that the oscillationmay be triggered by the sliding (P-sliding) necessary for P-bendformation (Shingyoji and Takahashi, 1995). These studiessuggest that if we use quiescent flagella instead of axonemalfragments for the study of the effects of imposed bending (Moritaand Shingyoji, 2004), we could define the direction of imposedbending by referring to the direction of the P-bend at the base(Fig. 1E). Furthermore, the second problem mentioned abovecan also be solved by using quiescent flagella. The dyneinactivity on doublet 3 is decreased at high Ca2+. Under such acondition, the velocity of microtubule sliding on doublet 3 islower than that on doublet 7 (Nakano et al., 2003). If this is alsotrue in the quiescent flagella, the sliding velocity can be a goodmarker to determine on which side of the CP dynein arms areactive.

In this study, we have examined whether bending itself or thedirection of bending is important for switching the dynein activityand how the combination of proximal and distal bends along thelength of the flagellum is associated with the switchingmechanism. By using elastase-treated Ca2+-induced quiescentflagella of sea urchin sperm, we show that a reversal of the slidingdirection is induced depending on the direction of imposedbending. During this study, we found that the presence or theabsence of a P-bend at the base of flagella appears to be closelyrelated to the regulation of the dynein activity by imposedbending. Therefore, we used three types of flagella with or withouta P-bend at the base [type 2 and type 3 (with), and type 1 (without);Fig. 1E]. In the type 1 experiments, quiescent flagella were cutat the base before inducing microtubule sliding (P-sliding). In thetype 2 experiments, quiescent flagella without cutting were used.In the type 3 experiments, quiescent flagella were cut at the baseafter microtubule sliding into a pair of bundles (P-sliding) wasinduced. We finally analyzed the velocity of sliding in theaxonemes under imposed bending and demonstrated that thevelocity of active backward sliding was lower than that of forwardsliding. Our results show that the switching of the dynein activityin flagella, which is induced by the combination of the directionof bending and the conformational state of the dynein-microtubuleinteraction, is the basis for flagellar oscillation.

ResultsEffects of the direction of imposed bending on the direction ofmicrotubule sliding in elastase-treated quiescent flagellasevered at the basal P-bendWe obtained quiescent flagella from demembranated sperm of thesea urchin in the presence of 10–4 M Ca2+ and 1 mM ATP and treatedthem with elastase after removal of ATP. When the elastase-treatedquiescent flagellum was cut with a glass microneedle at a distalregion of the basal P-bend and a small amount of ATP was appliedby 60 msecond UV-photolysis of 1 mM caged ATP, the part of theflagellum distal to the cut showed sliding disintegration (type 1 inFig. 1E). Among the axonemes that showed microtubule sliding,splitting into two microtubules bundles of unequal thickness wasdominant, and in 30-65% of such split axonemes the thinner of thetwo bundles slid towards the head (Fig. 2A,B; images 1-3). Thistype of microtubule sliding is caused by the so-called P-slidingowing to the activity of dynein arms on doublet 7 (Nakano et al.,2003). Successive applications of ATP by repetitive UV flashesinduced further sliding of the thinner bundle in the same forwarddirection.

Our previous study (Morita and Shingyoji, 2004) hasdemonstrated that in the elastase-treated axonemal fragments thatsplit into two bundles, a reversal of the direction of sliding can beinduced between the bundles by bending the region of overlapbetween the two bundles by more than 90° (>1.57 rad). We repeatedthis experiment under the same conditions (Fig. 1D) and confirmedthat bending induced the reversal of the direction of microtubulesliding in about half (48%) of the fragmented axonemes studied(n=23). By bending for 1.6-3.3 rad (supplementary material Fig.S1A), we tested the effect of the direction of imposed bending onthe direction of sliding between the doublet bundles obtained fromelastase-treated quiescent sperm flagella cut at the basal P-bend (type1 in Fig. 1E). Fig. 2 summarizes typical responses. In this study,we bent the axoneme in the plane as close to the plane of beat aspossible. The bending direction can be described as follows: whenthe region of overlap between the two bundles is bent in the samedirection as that of the original basal P-bend, it is categorized asbending in the P-bend direction (left panels in Fig. 2A,B); however,when the region of overlap is bent in the opposite direction relativeto the basal bend, which makes a bend similar to a reverse bend ina more distal region, it is categorized as bending in the R-benddirection (right panels in Fig. 2A,B).

When we bent the distal part of the region of overlap in the P-bend direction, subsequent application of ATP induced backwardsliding of the thinner bundle (Fig. 2A, left panels); such slidingoccurred in 56% of the axonemes studied (Fig. 2C, part A, whitebox). By contrast, when we bent the distal part of the region ofoverlap in the R-bend direction, photoreleasing ATP inducedforward sliding of the thinner bundle without a change in the slidingdirection (Fig. 2A, right panels) in most of the axonemes (82%)and backward sliding was induced in only one out of 22 axonemes(Fig. 2C, part A). Bending the proximal part of the overlap regionalso induced similar responses; backward sliding of the thinnerbundle occurred when the proximal region of overlap was bent inthe P-bend direction (Fig. 2B, left panels). This type of reversalwas observed in 43% of the axonemes (Fig. 2C, part B), whereasbending the proximal region in the R-bend direction did not inducebackward sliding (Fig. 2B, right panels; and Fig. 2C, part B). Theseresults show that the backward sliding can be induced by bendingin the P-bend direction those flagella that displayed the forwardsliding of the thinner bundle (P-sliding) before the bending.

Journal of Cell Science 121 (17)

Jour

nal o

f Cel

l Sci

ence

2835Switching of dynein activity in flagella

Detailed analysis of the backward sliding revealed that in fiveout of the 10 flagella bent in the distal region, backward slidingoccurred along the whole length of the region of overlap betweenthe two bundles (Fig. 2A, left), although in four of the 10 axonemes,the movement of the thinner bundle at the more distal part of thebending region was inhibited because of splitting of the thinnerbundle from the thicker bundle as shown in Fig. 2D. Sliding of thislatter case occurred when the distal edge of the thinner bundle wasin the bending region. By contrast, when the proximal region wasbent, as is shown in Fig. 2B (left) and Fig. 2E (the same flagellum

as shown in Fig. 2B, left), the backward sliding occurred only inthe proximal part of the bending region (open arrows), whereas theforward sliding continued in the distal part of the bending region(filled arrows). The simultaneous sliding in the opposite directionsat the proximal and distal parts of the imposed bending was observedin two of the three axonemes that showed backward sliding onbending the proximal region of the flagella.

In the type 1 experiments (Fig. 1E), the frequency of occurrenceof backward sliding induced by bending did not change comparedwith that in the experiments using fragmented axonemes (Fig. 1D).

Fig. 1. The regulation of dynein activity in a switching model of beating flagella and the experimental design for testing the hypothesis. (A) In the sea urchin spermflagellum, principal and reverse bends (PB and RB) are cyclically formed in the plane of beat (white), which is perpendicular to the plane of the CP (grey). (B) Theformation of PB and RB is due to P-sliding (PS) and R-sliding (RS), respectively, induced by the dynein arms of the doublets 7 and 3 (or 4) (Nakano et al., 2003).(C) A model, based on our previous study (Morita and Shingyoji, 2004), illustrating the postulated sequential regulation of dynein activity (with time and along theflagellum), which induces oscillation for a half beat cycle (from top to bottom). (D) Interpretation of the previous experiment (Morita and Shingyoji, 2004).Externally applied bending induces backward sliding between the two sets of doublets in the elastase-treated axonemal fragment when ATP was applied locally andtransiently by using caged-ATP with a UV flash. The testing hypothesis is that backward sliding is caused by switching of the activity of dynein from doublet 7 todoublet 3 (in the 8-3 pattern). (E) Overview of the present experiments. The effects of the direction of externally applied bending and those of the dyneinattachment states on the subsequent regulation of dynein activity are analysed by using three types (1-3) of axonemes obtained from elastase-treated quiescentflagella. Finally, the switching of dynein activity by bending is tested by measuring the velocity of microtubule sliding.

Jour

nal o

f Cel

l Sci

ence

2836 Journal of Cell Science 121 (17)

Fig. 2. Effects of imposed bending on the reversal of microtubule sliding in elastase-treated quiescent flagellar axonemes severed distally to the basal P-bend. (A,B) Video images with explanatory diagrams. UV flashes induced splitting of the flagellum into two doublet bundles. When the region of overlap was bent in theP-bend direction (left panels), the subsequent UV flashes induced backward sliding of the thinner bundle (open arrow in 5) in the whole region of the bending,whereas by imposed bending in the R-bend direction (right panels), forward sliding (filled arrows in 5) was induced. (C) Relative frequency of sliding patternsinduced by imposed bending. (D) Tracings showing that, by bending the distal region of the flagellum in the P-bend direction, backward sliding in the proximalregion of the bending (open arrow) and splitting of the distal edge of the thinner bundle in the bent region were induced by a UV flash. (E) Tracings taken from thesame axoneme as shown in B (left) after bending.

Jour

nal o

f Cel

l Sci

ence

2837Switching of dynein activity in flagella

However, the rate of bent axonemes that disintegrated into manysmaller bundles or individual doublets, which was about 40% in theprevious study (Morita and Shingyoji, 2004), decreased to 14% (boxeswith oblique lines in Fig. 2C). This shows that, in the type 1experiments, bending induced stable sliding between the two bundles,possibly as it was applied to the axoneme in the plane of the beat.

Effects of the direction of imposed bending on the direction ofmicrotubule sliding in elastase-treated quiescent flagellaThe above results show that the directionally controlled bending iseffective to induce stable sliding between the two bundles. Althoughthe effect may involve a reversal of the direction of sliding, the rateof reversals was not high (about 50%). This may mean that a specificdynein-regulating factor related to the switching mechanism maynot have been well controlled. The bending angle and the positionof bending along the axoneme appeared not to be important

(supplementary material Fig. S1). One strong candidate for thisswitching would be the conformational state of dynein-microtubuleinteraction under the control of the CP/RS system. In other words,the dynein-microtubule attachment states that are involved in theformation of bends would be related to the subsequent activity ofdynein arms located close to or within the bends. To test whetherthe conformational state of dynein arms along the axoneme affectsthe rate of backward sliding, we examined the effects of theexistence of the basal P-bend (type 2 and type 3 in Fig. 1E). Whenwe induced sliding in the elastase-treated axonemes of quiescentflagella without cutting at the basal P-bend (type 2 in Fig. 1E), byphotoreleasing ATP (Fig. 3A), 40-50% of the axonemes showedsliding in which the thinner bundles slid mostly towards the headwithout a large change in the shape of the basal P-bend. In theseaxonemes, forward sliding of the thinner bundles was inducedseveral times by repeating the UV flashes.

Fig. 3. Effects of imposed bending on the reversal of microtubule sliding in elastase-treated quiescent flagellar axonemes. (A-C) Video images with explanatorydiagrams. UV flashes induced splitting of the flagellum into two doublet bundles at the basal P-bend (A). Bending the distal part of the region of overlap of twobundles in the P-bend direction (B) induced further forward sliding (UV2), whereas bending in the R-bend direction (C) induced backward sliding (UV3, 4)(supplementary material Movies 1 and 2). (D) Relative frequency of sliding patterns induced by imposed bending. (E) Tracings showing that the effect of thepresence of the original basal P-bend on the direction of sliding. When the basal end of the flagellum was severed (Cut) after induction of the forward sliding at thebasal P-bend, and the proximal region was bent to induce a new R-bend in the more distal region of the P-bend (Bending), subsequent application of ATP (UV3)induced backward sliding (open arrows) only at the proximal part of the imposed bending. However, the distal part still continued forward sliding (filled arrows).

Jour

nal o

f Cel

l Sci

ence

2838

When we bent the distal part of the region of overlap in the P-bend direction (Fig. 3B), the sliding direction of the thinner bundledid not change and forward sliding was induced by UV flashes in63% of the flagella (Fig. 3D; n=16). When we bent the distal partof the region of overlap in the R-bend direction, however,subsequent application of ATP induced backward sliding of thethinner bundles in 71% of the axonemes (Fig. 3C,D; n=31), andthe forward sliding decreased to 6.5%. The backward sliding wasinduced along the whole region of overlap (Fig. 3C) in 19 out ofthe 22 axonemes; the three that were not induced showed splittingof the distal part of the thinner bundle near the bent region. Theseresponses resembled those observed in the quiescent flagella witha cut at the P-bend described above (type 1 in Fig. 1E; Fig. 2),although the effective direction of bending to induce the backwardsliding was opposite and the frequency of the occurrence of thebackward sliding was different. The rate of disintegration intosmaller bundles or individual doublets was also about 18% in thetype 2 experiments, which was similar to the rate in the type 1experiments.

If the presence of the basal P-bend functions to maintain theconformational state of dynein-microtubule interaction through theCP/RS system, the flagellum cut at the proximal but not at the distalpart of the P-bend would be expected to preserve the conformationalinformation and to respond to imposed bending (type 3 in Fig. 1E)(in the same way as the flagellum without cutting). Fig. 3E showssequential tracings of one example. In this experiment P-slidingwas first induced at the basal P-bend (UV1, 2) and then the flagellumwas cut at the base (Cut). By manipulating only the proximal regionof the flagellum, a new R-bend was formed in a more distal regionof the basal P-bend (Bending). ATP application (UV3) inducedbackward sliding in a proximal region (open arrows). We foundthat in the distal part of the bending region, forward sliding (P-sliding) still occurred (filled arrows). In two other quiescent flagellacut at the base after forward sliding was induced at the original P-bend, backward sliding was induced by bending in the R-benddirection. These results indicate that, even in the flagellum cut atthe base, the information provided by the dynein–microtubuleinteraction at the basal P-bend seems to be maintained, unless furthermicrotubule sliding is induced after cutting.

Velocity of microtubule sliding on thinner and thicker bundlesin quiescent flagellaIn axonemal fragments, the activity of dynein on doublet 3 isinhibited at 10–4 M Ca2+, whereas that on doublet 7 is not (Nakanoet al., 2003). To test whether this is also true in the quiescent flagella,we analyzed the behaviour of microtubules interacting with thedynein arms exposed on the thicker or the thinner bundle obtainedfrom elastase-treated axonemes.

Quiescent flagella treated with elastase also showed splitting intotwo doublet bundles in the presence of 1 mM ATP and 10–4 M Ca2+.In some cases, sliding brought about splitting along the whole lengthof a flagellum (Fig. 4A), in which the thinner bundle slid towardthe head. After the ATP was replaced with 1 mM caged-ATP, weapplied singlet microtubules to the dynein arms exposed on thethinner and thicker bundles, and analyzed their movement.Photoreleasing ATP from caged-ATP induced microtubule slidingon each bundle and all the observed microtubules slid away fromthe head of the flagella (Fig. 4A), indicating that the polarity ofdynein force generation on these bundles is towards the minus-endof the microtubule. Fig. 4B shows distribution of the velocity ofmicrotubule sliding on the thinner (left) and thicker (right) bundles.

The velocity on the thinner bundle was higher than that on thethicker bundle at 10–4 M Ca2+. The difference was statisticallysignificant (Mann-Whitney U test, P<0.05). These results showedthat in quiescent flagella, the activity of dynein arms on doublet 3is inhibited at a high Ca2+ concentration, and the inhibition can bedetected as a decrease in the velocity of microtubule sliding.

Switching of the dynein activity at the backward slidingIf imposed bending can trigger switching of the dynein activitybetween the two sides of the CP, the backward sliding induced bythe imposed bending in elastase-treated quiescent flagella shouldbe caused by the activity of dynein arms on the thicker bundle. Andif this is the case, the velocity of the backward sliding is expectedto be lower than that of forward sliding at 10–4 M Ca2+. To test this,we measured the sliding velocity of forward and backward slidinginduced in elastase-treated axonemes. We used three types ofelastase-treated flagella: axonemal fragments, quiescent flagellasevered at the P-bend and quiescent flagella without severing. Wecompared the velocity of forward sliding with or without bending,and the velocity of backward sliding induced by bending (Fig. 4C).As the velocity of microtubule sliding induced by photoreleasingATP from caged-ATP was almost constant from 0.2 to 1.5 secondsafter the UV flash, we determined the velocity within 0.2-1.3seconds of the UV flash. As we describe in the next section, someflagella showed delayed responses after the bending, so we choseonly the flagella that immediately responded to the UV flash. Thevelocities of the forward sliding after bending tended to be higherthan those without bending, which is consistent with the previousreport (Morita and Shingyoji, 2004) indicating that the imposedbending increases the dynein activity.

Comparing the velocities of backward and forward sliding, wefound that under imposed bending the velocity of the backwardsliding was significantly lower than that of the forward sliding. Thedecrease of sliding velocity in backward sliding is statisticallysignificant in any of the three types of flagella (P<0.05 for axonemalfragments and severed quiescent flagella; P<0.01 for quiescentflagella). The results support the idea that when the direction ofsliding changes from the forward sliding (P-sliding) to the backwardsliding, the site of dynein activity changes from the thinner to thethicker bundle and the backward sliding is caused by the activityof the dynein arms of the thicker bundle. More specifically, theyindicate that imposed bending induces switching of the dyneinactivity from doublet 7 to doublet 3.

Time lag at the onset of backward sliding Throughout the above experiments, we found that the forwardsliding always occurred immediately after an application of a UVflash. Such immediate response in the forward sliding wasindependent of imposed bending, in quiescent flagella that werecut at the basal P-bend as well as those without a cut. Fig. 5A-Eshows examples of time courses of the immediate response recordedin the absence (A and UV1 in B-E) and presence (UV2, 3 in B) ofimposed bending. Similar immediate response to a UV flash wasalso observed in most of the backward sliding (Fig. 5C, UV2-4).However, we found that some of the backward sliding occurredafter a short delay (0.8-1.3 seconds) (Fig. 5D, UV2) or a pause(Fig. 5E, UV2). In such flagella, immediate backward sliding wasinduced at the next UV flash (UV3 in Fig. 5D,E).

About 27% and 47% of the axonemes both severed at the basalP-bend and unsevered, respectively, showed backward sliding witha time lag that included a short delay and a pause (Fig. 5F, Bw1

Journal of Cell Science 121 (17)

Jour

nal o

f Cel

l Sci

ence

2839Switching of dynein activity in flagella

and Bw2). The time lag was also observed in the backward slidingin axonemal fragments at a smaller rate (8%). Forward sliding inany type of the axonemes never had a time lag (Fig. 5F, Fw). It isinteresting that once backward sliding was induced, no delay wasobserved at the onset of the second backward sliding induced bythe next UV flash (Fig. 5D, UV3). The average velocity of thebackward sliding that followed the pause (2.3±1.1 μm/second, n=5)was significantly lower than that of the forward sliding (6.5±3.0μm/second, n=23) after imposed bending in the quiescent flagellawith and without a cut. The results indicate that the time lag at theonset of the backward sliding is probably associated with themechanism of switching of dynein activity between both sides ofthe CP.

DiscussionBending-induced switching of dynein activity as the basis forflagellar oscillationAs the regulation of flagellar motility is basically coupled with themechanical bending itself (Hayashibe et al., 1997; Holcomb-Wygleet al., 1999; Ishikawa and Shingyoji, 2007; Morita and Shingyoji,2004; Okuno and Hiramoto, 1976; Shingyoji et al., 1991), in orderto understand the mechanism of flagellar oscillation, an analysis of

the effects of bending on microtubule sliding seems indispensable.The switching hypothesis (Satir, 1985), proposed as a pioneeringmodel to explain the regulation of dynein activity in beating flagellaand cilia, postulates alternate activation of the dynein arms on eitherhalf of the axoneme beside the CP. Several ultrastructuralobservations demonstrating splitting of the axoneme into twodoublet bundles suggest that the switching of active microtubulesliding occurs between the two sides of the CP during flagellarbeating (Fig. 1A,B) (Holwill and Satir, 1994; Nakano et al., 2003;Satir and Matsuoka, 1989; Wargo et al., 2004). However, there hasbeen no clear evidence to demonstrate which dynein arms are activeand how the activity of dynein is switched on and off when flagellabeat. To demonstrate this, detailed analysis of microtubule slidingis necessary in axonemes that are capable of beating.

A previous study (Morita and Shingyoji, 2004) has demonstratedthat a reversal in the direction of microtubule sliding between splitbundles of the elastase-treated axoneme of sea urchin spermflagella, which retain the ability of beating, is induced by imposedbending of the region of overlap of the two bundles. As themicrotubule sliding that splits the axoneme into two bundles iscaused mainly by the activity of dynein on doublet 7 under high-ATP and high-Ca2+ conditions, the reversal of the sliding direction

Fig. 4. Velocity of microtubule sliding and sliding disintegration in elastase-treated flagella. (A) Movie images with tracings (top panels) showing sliding of singletmicrotubules interacted with dynein arms exposed on the thinner (left panels) and the thicker (right panels) bundles of doublets obtained by splitting between thebundles in quiescent flagella. The microtubules moved away from the head after a UV flash. (B) Distribution of sliding velocities of singlet microtubules on thethinner (left) and thicker (right) doublet bundles at 1 mM caged-ATP and 10–4 M Ca2+. The velocity on the thicker bundles was significantly lower than that on thethinner bundles. (C) Average sliding velocities with their standard deviations (bars) measured in forward (FW) and backward (BW) sliding with and withoutimposed bending in axonemal fragments (left graph), in the quiescent flagella severed at the basal P-bend (middle graph) and in the quiescent flagella (right graph).Asterisks indicate that the differences between the two are statistically significant (Mann-Whitney U test, **P<0.01; *P<0.05).

Jour

nal o

f Cel

l Sci

ence

2840

is probably due to the activity of dynein on doublet 3 (or 4) unlessthe polarity of dynein force generation changes. But we did notknow whether this was really the case (Fig. 1D).

In the present study, we used the quiescent flagellum with thesperm head intact and microtubule sliding was induced in thequiescent axonemes after elastase treatment. This procedure enabledus to determine the direction of beating plane in addition to thelongitudinal polarity, proximal and distal directions of the flagellumthat correspond to the minus and plus ends of the microtubule. Inthis way, when the thinner of the two bundles slides towards thehead, we can infer that the dynein arms on doublet 7 are the maingenerator of power for the sliding (Nakano et al., 2003) (Fig. 1E).If the activity of dynein arms is switched by imposed bending fromdoublet 7 to doublet 3 on the thicker bundle, then the direction ofsliding of the thinner bundle is expected to reverse.

In sea urchin sperm, the dynein activity on both sides of the CPis differently controlled and the dynein arms on doublet 3, whichare near the C2 microtubule of the CP, are less active than thoseon doublet 7 (Nakano et al., 2003). This study shows that this featureof lower sliding velocity is also observed in the quiescent flagella

(Fig. 4B). By comparing the sliding velocities of forward andbackward sliding, we demonstrated that the backward sliding wassignificantly slower than the forward sliding, indicating that thebackward sliding was induced by the activity of dynein on doublet3. Thus, the present results show that the reversal of the directionof microtubule sliding is caused by switching of dynein activitybetween the two sides of the CP. This is the basis for the alternationof microtubule sliding underlying the cyclical bending of flagella.

Factors important for switching the dynein activityThe amount of shear and the curvature of the axoneme are variedby imposed bending. Backward sliding was induced by bendingwith a curvature of 0.22-0.60/μm when the bending angle was largerthan 1.7 rad [except in one case (0.81 rad)] in all the axonemesused in this study. However, there was no correlation between thebending angle and the occurrence of backward sliding(supplementary material Fig. S1), indicating that the absolute valueof bending angle, which corresponds to the amount of shear, is notthe sole determinant of the switching of dynein activity. The mostimportant factors associated with the switching of dynein activity

are the direction of bending and theconformational state of the dynein–microtubuleinteraction.

In the elastase-treated quiescent flagella thathad been severed near the base, bending theregion of overlap between the two bundles inthe same direction as that of the originalprincipal bend at the base (the P-bend direction)induced backward sliding of the thinner bundle,but bending in the R-bend direction did notinduce backward sliding (Fig. 2). The resultsshowed that bending was effective only whenit was applied in a fixed direction. Thus, bendingin the P-bend direction and that in the R-benddirection did not equally affect the regulationof dynein activity. The present results show thatthe dynein arms (mainly those on doublet 7) thatgenerate principal sliding (Ps) in the axonemerespond to bending in the P-bend direction andthat their activity can be switched off, which isfollowed by activation of the dynein arms onthe opposite side of the CP to induce reversesliding (Rs) (Fig. 6C1). This type of regulation,which is also explained according to curvature-controlled models (Fig. 6A) (Brokaw, 1985;Brokaw, 2001), would be important for theinitiation of cyclical bending by switching ofthe dynein activity at the flagellar base. Bycontrast, the present observation showing thatbending the axoneme in the R-bend direction

Journal of Cell Science 121 (17)

Fig. 5. Profiles of sliding disintegration between twodoublet bundles in elastase-treated quiescent flagellawith (A-D) or without (E) severing at the basal P-bend.The distance of sliding of the thinner bundle towardsthe base was plotted against time. B and D correspondto Fig. 2A, right and left, respectively. (F) Relativefrequencies of time lag at the onset of sliding with orwithout imposed bending in the axonemal fragments(upper three boxes) and the quiescent flagella with orwithout severing at the basal P-bend (lower fourboxes).

Jour

nal o

f Cel

l Sci

ence

2841Switching of dynein activity in flagella

cannot induce a reversal of the sliding direction is not consistentwith the models controlled by the curvature (Fig. 6A). The presentresults and conclusions apply directly only to the case of a newreverse sliding following a preceding P-sliding. However, we haveno reason to believe that the situation would be different for a newP-sliding in the axoneme displaying an R-sliding, if the technicallimitations preventing its observation could be overcome.

In the elastase-treated quiescent flagella (Fig. 3), the effectivedirection of bending to induce backward sliding was reversed.Bending in the R-bend direction in a region more distal to theoriginal P-bend induced an R-sliding by switching of the dyneinactivity. In this case, the formation of a pair of bends composed ofa proximal P-bend and a more distal R-bend activated the dyneinarms to produce reverse sliding along the bend. The results indicatethat the difference in the amount of sliding induced by a pair ofopposite bends is important for the switching of dynein activity.The localized cyclical bend formation (Fig. 6B) induced bysuccessive iontophoretic application of ATP to either a protease-untreated or an elastase-treated sea urchin sperm flagellar axonemes(Shingyoji et al., 1977; Shingyoji and Takahashi, 1995) probablyinvolves a regulatory mechanism similar to that underlying theactivation of dynein by paired bends. In these locally reactivatedaxonemes, microtubule sliding occurs only in the region (interbendregion) between the pair of opposite bends, and a reversal of the

sliding direction between the two bends leads to alternation of thedirection of these bends (Fig. 6B). By contrast, the backward slidingobserved in the present study was induced either along the entireflagellum, including the pair of bends (Fig. 3C) or the proximalpart of the bending region (Fig. 3E), and such a change in the slidingdirection caused by bending may be involved in the regulation ofmicrotubule sliding underlying the propagation (Fig. 6C2) andgrowth (Fig. 6C3) of bends in normal beating. Thus, coupling ofthe mechanism regulating microtubule sliding in the paired bendformation (Fig. 6C2,3) with the regulation that induces a reversalof the sliding direction in the bending region (Fig. 6C1,2) mayconstitute a basis for the coordination of cyclical bend formationand propagation. This idea is supported by the previous finding thatcyclical bend formation and propagation can be induced by imposedbending so as to form a pair of opposite bends in demembranatednonmotile sperm flagella at 2-3 μM ATP (Ishikawa and Shingyoji,2007). The influence of the distal bend to the proximal sliding isalso indicated by the previous study showing that immotile bullsperm flagellum in the presence of Ni2+ recovers its beating bybending the distal region in the direction opposite to the originalbasal bend (Lindemann et al., 1995). These results suggest that theconformational state of dynein-microtubule interaction in thecombination of the proximal and distal bends is responsible for thecontinuous switching of the dynein activity along the flagellum.

In beating flagella, the regulation of the activity of all dyneinarms in the bend plane as well as the switching of dynein activityin the plane of the CP in the interbend regions contribute to producecyclical beating (Fig. 1C). To understand the flagellar oscillation,we have to elucidate the mechanism regulating the combination ofthese dynein functions. The disintegration of axonemes into smallerbundles or individual doublets induced by bending, which wasobserved frequently in the previous study (Morita and Shingyoji,2004), was infrequent in the present study. This suggests that theactivation of all dynein arms by the imposed bending requires thatthe force be applied to dynein arms in the direction not parallel tothe beat plane. In addition to the change in the dynein attachmentstate resulting from the shear induced by the imposed bending,axonemal distortion would occur along the doublets in the bendplane, tilting the radial spokes and changing the geometricrelationship between the doublets (Lindemann, 1994a; Lindemanand Mitchell, 2007). These factors may cause activation of thedynein arms on the doublets in the bend plane, and are probablylinked to the switching of the activity of dynein on both sides ofthe CP in the manner predicted by the ‘geometric clutch’ model(Lindemann, 1994a; Lindemann, 1994b). Thus, the combination ofthe regulation of the activity of all dynein arms in the bend planewith the mechanism of switching the dynein activity on both sidesof the CP is essential for flagellar beating.

Roles of the CP/RS system and bend asymmetryThe change in the CP/RS interaction with the direction of bendingprobably mediates the mechanical feedback from flagellar bending(Bannai et al., 2000; Mitchell and Nakatsugawa, 2004; Warner andSatir, 1974). The regulation through chemical as well as mechanicalsignals may also play an important role in this process. Among thechemical signals, regulation of dynein activity by proteinphosphorylation and dephosphorylation seems important (Inaba,2002; Nakajima et al., 2005; Smith and Yang, 2004) although theexact pathways of this regulation remain unclear (Yoshimura et al.,2007). The regulation through the CP/RS system is responsible ata physiological, high-ATP condition. By contrast, at lower ATP

Fig. 6. Relative direction of shear forces or microtubule sliding within theaxoneme during bend formation. (A) The basis for curvature controlledmodels for flagella by Brokaw (Brokaw, 2001) with some modifications. Theblack arrows indicate the direction of shear forces or of sliding in theflagellum. For bends propagating from left to right, the direction of internalsliding must reverse at the points shown by the open arrows when thecurvature exceeds critical curvatures. (B) Localized cyclical bending inducedby iontophoretic applications of ATP to a demembranated or to ademembranated and elastase-treated sea urchin sperm flagellum (Shingyojiand Takahashi, 1995). (C) Interpretation of the backward sliding observed inthe present study. (1) In the elastase-treated quiescent flagella severed at thebase (Fig. 2A,B, left), P-sliding is induced by ATP application (left). Bendingin the P-bend direction (middle) induces backward sliding (right) mainly in theregion proximal to the bending. (2,3), In the elastase-treated quiescentaxoneme (Fig. 3C,E), P-sliding occurs in the distal part of the basal P-bend(left). Bending the distal region of the flagellum in the R-bend direction(middle) induces backward sliding in the whole region of the bending (2,right) or only in the proximal region of the bending (3).

Jour

nal o

f Cel

l Sci

ence

2842

concentrations (less than 0.1 mM), the presence of CP/RS systemis not essential (Bannai et al., 2000; Nakano et al., 2003; Yoshimuraet al., 2007), suggesting that other inherent asymmetries in theaxoneme are required. The role of the CP/RS system at high ATP,which, as is discussed below, is probably associated with theinhibition and activation of the dynein activity by ATP and ADP,respectively (Yoshimura et al., 2007), is similar in cilia and flagellaof different species (Bannai et al., 2000). The recent finding of thecentral pair component hydin, the gene encoding which is presentbroadly in organisms with the ability to assemble motile 9+2axonemes, may indicate that switching of dynein activity throughthe CP/RS system is a conserved property of axonemal mechanismsfor alternating bends (Lechtreck and Witman, 2007; Lechtreck etal., 2008).

In the present study, we observed a time lag at the onset ofbackward sliding. This may reflect a process in the transduction ofchemical signals from CP1 to CP2 in the region of bending. Innormal flagellar beating, bending is formed in alternate directionsby rapid and smooth switching of the dynein activity without delay.In the present experiments, dynein activation was brought about bya UV-photolysis of caged-ATP, which may have modified the signaltransduction process and enabled us to observe the event at theswitching process. The time lag occurred more frequently in theelastase-treated quiescent flagella than in the other two types ofaxonemes (Fig. 5F). It is tempting to speculate that this may berelated to the differences in propagation ability among the threetypes of axonemes. In the models proposed by Machin and Brokaw(Machin, 1958; Brokaw, 1971; Brokaw, 1985) to explain theflagellar movement, wave propagation can occur if a time lag isintroduced into the relationship between the active bending momentand the curvature. In our study, only backward sliding observed inthe elastase-treated quiescent flagella, which appeared to retain theregulatory mechanism related to bend propagation (Fig. 6C2),accompanied a time lag. This may indicate a close associationbetween the time lag and wave propagation. A similar delay betweenthe formation of bends was also reported in reactivated Ciona spermflagella under high-LiCl (10 mM) and low-MgATP (10 μM)conditions (Brokaw, 1989b), in which a pause appears after the P-bend growth. At the end of the pause, active R-sliding is turned onthroughout the P-bend and this active sliding causes the P-bend tostart propagating. This also supports our idea that the backwardsliding (R-sliding) induced throughout the pair of bends (Fig. 6C2)is related to the regulation of sliding that produces wave propagation.

Our recent studies have revealed that the key mechanisms of theregulation of the activity of flagellar dynein are ATP-inducedinhibition and ADP-induced activation. These mechanisms areinvolved in the regulation of both the dynein molecules (Inoue andShingyoji, 2007) and flagellar beating (Yoshimura et al., 2007). Theinhibition and activation of dynein activity seem to be associatedwith binding of ATP only and both ATP and ADP, respectively, tothe three noncatalytic regulatory sites of dynein (Inoue andShingyoji, 2007). We have shown that caged-ATP behaves in thesame way as a non-hydrolysable ATP analogue and binds stably todynein (Inoue and Shingyoji, 2007). In the presence of caged-ATP,dynein is in an inhibited state and after hydrolysis of the UV-releasedATP the dynein becomes active probably by binding of ADP tosome of the regulatory sites. According to the protocol forreactivating the dynein by UV photolysis of caged-ATP, the dyneinregulatory sites, as well as the ATP hydrolysis site, were in aninactive state before the UV flash and the dynein was forming acrossbridge between doublet microtubules, which may be important

for maintaining the conformational states of dynein-microtubuleinteraction along the flagellum.

It has recently been shown that ADP releases the inhibition offlagellar movement by ATP, and a part of the activation of dyneinthrough protein phosphorylation appears to be caused by the ADPbinding to dynein (Yoshimura et al., 2007). Our preliminaryexperiments show that the induction of backward sliding byimposed bending requires the presence of ADP, suggesting that theactivation of dynein by mechanical force is also induced by amechanism involving ADP binding. Mechanical deformation of theaxoneme caused by bending may induce CP/RS-mediatedphosphorylation and/or dephosphorylation of some axonemalproteins. Such regulation of the dynein arms on the two sides ofthe CP is probably caused by some direct (chemical and mechanical)influences from the CP1 and CP2, although chemical and structuraldifferences between the CP1 and CP2, which have been reportedin Chlamydomonas (Smith and Yang, 2004), have not beendemonstrated in sea urchin sperm. By contrast, the activity of dyneinarms on other doublets may be regulated by the mechanicaldeformation but would be independent of the CP/RS system. How,for oscillatory bending, the mechanical and chemical signalsmodulate and coordinate the dynein activity by ADP binding to thedynein regulatory sites, and how these signals relate to each otherare important points for the future study.

Materials and MethodsPreparation of axonemesMicrotubule sliding of elastase-treated flagellar axonemes of spermatozoa of the seaurchins Anthocidaris crassispina and Pseudocentrotus depressus was studiedaccording to the previous method (Morita and Shingyoji, 2004) with somemodifications. The demembranated sperm were labelled with 9.4 μMtetramethylrhodamine (Molecular Probes C-1171) for 8 minutes on ice for theexperiments of imposed bending (for Figs 2, 3, 5; Fig. 4C) or labelled with 13 μMcy5 (Amersham Biosciences PA25001) for the experiments of sliding movement ofrhodamine-labelled singlet microtubules on doublet bundles of the axonemes (Fig.4A,B) (Nakano et al., 2003). The sliding disintegration was induced in the assaysolution containing 20 mM HEPES-KOH (pH 7.8) instead of Tris-HCl (Nakano etal., 2003).

Observation of sliding and application of bendingA quiescent waveform was induced in the reactivating solution (HEPES) containing10–4 M Ca2+ and 1 mM ATP (for A. crassispina and for P. depressus in Fig. 4A,B)or 5 mM ATP (for P. depressus in the remaining experiments). A suspension of thequiescent flagella was introduced into a 10 μl perfusion chamber (Morita andShingyoji, 2004). ATP concentration was decreased by perfusing 40 μl of reactivatingsolution (HEPES) without ATP and with 20 units/ml hexokinase and 20 mM glucosemore than three times. The quiescent flagella were treated with elastase and theirsliding disintegration was induced by photoreleased ATP from caged-ATP (p3-[1-(2-Nitrophenyl) ethyl] ATP, Dojindo 349-05501) according to a previously publishedmethod (Morita and Shingyoji, 2004), except that 1.0 mM caged-ATP with 40 units/mlhexokinase at 10–4 M (for A. crassispina and P. depressus) or 10–3 M (for P. depressus)Ca2+ were used. Axonemes were bent with a glass microneedle according to previousmethods (Morita and Shingyoji, 2004). The experiments of sliding of singletmicrotubules on doublet bundles were carried out according to a previous method(Nakano et al., 2003). Recording and analysis of the microtubule sliding were alsostudied according to previous methods (Morita and Shingyoji, 2004).

We would like to express our gratitude to Professor Keiichi Takahashifor discussion and improving of the manuscript. This work wassupported by Research Fellowships of the Japan Society for thePromotion of Science for Young Scientists (No. 17-12026 for S.H.)and Grant-in-Aid for Scientific Research on Priority Area from theMinistry of Education, Culture, Sports, Scientific and Technology, theJapanese Government (No. 16083203 for C.S.).

ReferencesBannai, H., Yoshimura, M., Takahashi, K. and Shingyoji, C. (2000). Calcium regulation

of microtubule sliding in reactivated sea urchin sperm flagella. J. Cell Sci. 113, 831-839.

Journal of Cell Science 121 (17)

Jour

nal o

f Cel

l Sci

ence

2843Switching of dynein activity in flagella

Brokaw, C. J. (1971). Bend propagation by a sliding filament model for flagella. J. Exp.Biol. 55, 289-304.

Brokaw, C. J. (1985). Computer simulation of flagellar movement. VI. Simple curvature-controlled models are incompletely specified. Biophys. J. 48, 633-642.

Brokaw, C. J. (1989a). Direct measurements of sliding between outer doublet microtubulesin swimming sperm flagella. Science 243, 1593-1596.

Brokaw, C. J. (1989b). Operation and regulation of the flagellar oscillator. In Cell Movement,vol. 1 (ed. F. D. Warner, P. Satir and I. R. Gibbons), pp. 267-279. New York: Alan R.Liss.

Brokaw, C. J. (2001). Simulating the effects of fluid viscosity on the behaviour of spermflagella. Math. Meth. Appl. Sci. 24, 1351-1365.

Eshel, D. and Gibbons, I. R. (1989). External mechanical control of the timing of bendinitiation in sea urchin sperm flagella. Cell Motil. Cytoskeleton 14, 416-423.

Hayashibe, K., Shingyoji, C. and Kamiya, R. (1997). Induction of temporary beating inparalyzed flagella of Chlamydomonas mutants by application of external force. CellMotil. Cytoskeleton. 37, 232-239.

Holcomb-Wygle, D. L., Schmitz, K. A. and Lindemann, C. B. (1999). Flagellar arrestbehavior predicted by the Geometric Clutch model is confirmed experimentally bymicromanipulation experiments on reactivated bull sperm. Cell Motil. Cytoskeleton 44,177-189.

Holwill, M. E. and Satir, P. (1994). Physical model of axonemal splitting. Cell Motil.Cytoskeleton 27, 287-298.

Inaba, K. (2002). Dephosphorylation of Tctex2-related dynein light chain by type 2A proteinphosphatase. Biochem. Biophys. Res. Commun. 297, 800-805.

Inoue, Y. and Shingyoji, C. (2007). The roles of noncatalytic ATP binding and ADP bindingin the regulation of dynein motile activity in flagella. Cell Motil. Cytoskeleton 64, 690-704.

Ishikawa, R. and Shingyoji, C. (2007). Induction of beating by imposed bending ormechanical pulse in demembranated, motionless sea urchin sperm flagella at very lowATP concentrations. Cell Struct. Funct. 32, 17-27.

Lechtreck, K. and Witman, G. B. (2007). Chlamydomonas reinhardtii hydin is a centralpair protein required for flagellar motility. J. Cell Biol. 176, 473-482.

Lechtreck, K., Delmotte, P., Robinson, M. L., Sanderson, M. J. and Witman, G. B.(2008). Mutations in Hydin impair ciliary motility in mice. J. Cell Biol. 180, 633-643.

Lindemann, C. B. (1994a). A “Geometric Clutch” hypothesis to explain oscillations ofthe axoneme of cilia and flagella. J. Theor. Biol. 168, 175-189.

Lindemann, C. B. (1994b). A model of flagellar and ciliary functioning which uses theforces transverse to the axoneme as the regulator of dynein activation. Cell Motil.Cytoskeleton 29, 141-154.

Lindemann, C. B. and Mitchell, D. R. (2007). Evidence for axonemal distortion duringthe flagellar beat of Chlamydomonas. Cell Motil. Cytoskeleton 64, 580-589.

Lindemann, C. B., Walker, J. M. and Kanous, K. S. (1995). Ni2+ inhibition inducesasymmetry in axonemal functioning and bend initiation of bull sperm. Cell Motil.Cytoskeleton 30, 8-16.

Machin, K. E. (1958). Wave propagation along flagella. J. Exp. Biol. 35, 796-806.Mitchell, D. R. and Nakatsugawa, M. (2004). Bend propagation drives central pair rotation

in Chlamydomonas reinhardtii flagella. J. Cell Biol. 166, 709-715.

Morita, Y. and Shingyoji, C. (2004). Effects of imposed bending on microtubule slidingin sperm flagella. Curr. Biol. 14, 2113-2118.

Nakajima, A., Morita, M., Takemura, A., Kamimura, S. and Okuno, M. (2005). Increasein intracellular pH induces phosphorylation of axonemal proteins for activation of flagellarmotility in starfish sperm. J. Exp. Biol. 208, 4411-4418.

Nakano, I., Kobayashi, T., Yoshimura, M. and Shingyoji, C. (2003). Central-pair-linkedregulation of microtubule sliding by calcium in flagellar axonemes. J. Cell Sci. 116,1627-1636.

Okuno, M. and Hiramoto, Y. (1976). Mechanical stimulation of starfish sperm flagella.J. Exp. Biol. 65, 401-413.

Omoto, C. K., Gibbons, I. R., Kamiya, R., Shingyoji, C., Takahashi, K. and Witman,G. B. (1999). Rotation of the central pair microtubules in eukaryotic flagella. Mol. Biol.Cell 10, 1-4.

Sale, W. S. (1985). Study of the properties of MgATP2–induced stationary bends indemembranated sea urchin sperm. Cell Motil. 5, 209-224.

Sale, W. S. (1986). The axonemal axis and Ca2+-induced asymmetry of active microtubulesliding in sea urchin sperm tails. J. Cell Biol. 102, 2042-2052.

Sale, W. S. and Satir, P. (1977). Direction of active sliding of microtubules in Tetrahymenacilia. Proc. Natl. Acad. Sci. USA 74, 2045-2049.

Satir, P. (1985). Switching mechanisms in the control of ciliary motility. Mod. Cell Biol.4, 1-46.

Satir, P. and Matsuoka, T. (1989). Splitting the ciliary axoneme: implications for a “switch-point” model of dynein arm activity in ciliary motion. Cell Motil. Cytoskeleton 14, 345-358.

Shingyoji, C. and Takahashi, K. (1995). Cyclical bending movements induced locally bysuccessive iontophoretic application of ATP to an elastase-treated flagellar axoneme. J.Cell Sci. 108, 1359-1369.

Shingyoji, C., Murakami, A. and Takahashi, K. (1977). Local reactivation of Triton-extracted flagella by iontophoretic application of ATP. Nature 265, 269-270.

Shingyoji, C., Gibbons, I. R., Murakami, A. and Takahashi, K. (1991). Effect of imposedhead vibration on the stability and waveform of flagellar beating in sea urchinspermatozoa. J. Exp. Biol. 156, 63-80.

Shingyoji, C., Yoshimura, K., Eshel, D., Takahashi, K. and Gibbons, I. R. (1995). Effectof beat frequency on the velocity of microtubule sliding in reactivated sea urchin spermflagella under imposed head vibration. J. Exp. Biol. 198, 645-653.

Smith, E. F. and Yang, P. (2004). The radial spokes and central apparatus: mechano-chemical transducers that regulate flagellar motility. Cell Motil. Cytoskeleton 57, 8-17.

Vale, R. D. and Toyoshima, Y. Y. (1988). Rotation and translocation of microtubules invitro induced by dyneins from Tetrahymena cilia. Cell 52, 459-469.

Wargo, M. J., McPeek, M. A. and Smith, E. F. (2004). Analysis of microtubule slidingpatterns in Chlamydomonas flagellar axonemes reveals dynein activity on specific doubletmicrotubules. J. Cell Sci. 117, 2533-2544.

Warner, F. D. and Satir, P. (1974). The structural basis of ciliary bend formation. Radialspoke positional changes accompanying microtubule sliding. J. Cell Biol. 63, 35-63.

Yoshimura, A., Nakano, I. and Shingyoji, C. (2007). Inhibition by ATP and activationby ADP in the regulation of flagellar movement in sea urchin sperm. Cell Motil.Cytoskeleton 64, 777-793.

Jour

nal o

f Cel

l Sci

ence